Download YouTube videos with AWS Lambda and store them on S3

Michael Wittig – 17 May 2019

Recently, I was faced with the challenge to download videos from YouTube and store them on S3.

Download YouTube videos with AWS Lambda

Sounds easy? Remember than Lambda comes with a few limitations:

  1. 512 MB of disk space available at /tmp
  2. 3008 MB of memory
  3. 15 minutes maximum execution time

While working on a solution, I encountered multiple problems:

  1. Download the video from YouTube to /tmp and then upload it to S3: Does not work with videos larger than 512 MB.
  2. Download the video from YouTube into memory and then upload it to S3: Does not work with videos larger than ~3 GB.
  3. Download the video from Youtube and stream it to S3 while downloading: Works for all videos that can be processed within 15 minutes. I have not found a video that took longer than a few minutes to process.

Let’s look at how I finally solved the problem with a streaming approach in Node.js. I use the youtube-dl library to get easy access to YouTube videos.

First, we create a PassThrough stream in Node.js. A pass-through stream is a duplex stream where you can write on one side and read on the other side.

const stream = require('stream');
const passtrough = new stream.PassThrough();

Next, we need to write data to the stream. This is done by the youtube-dl library.

const youtubedl = require('youtube-dl');
const dl = youtubedl(event.videoUrl, ['--format=best[ext=mp4]'], {maxBuffer: Infinity});
dl.pipe(passtrough); // write video to the pass-through stream

And finally, we need to upload the stream to S3. We make use of the Multipart Upload feature of S3 which allows us to upload a big file in smaller chunks. This way, we only have to buffer the small junk (64 MB in this case) in memory and not the whole file.

const AWS = require('aws-sdk');
const upload = new AWS.S3.ManagedUpload({
params: {
Bucket: process.env.BUCKET_NAME,
Key: 'video.mp4',
Body: passtrough
},
partSize: 1024 * 1024 * 64 // 64 MB in bytes
});
upload.send((err) => {
if (err) {
console.log('error', err);
} else {
console.log('done');
}
});

That’s it. Now you can download YouTube videos of any size with Lambda and upload them to S3. I recommend running the code in a “big” Lambda function with 3008 MB of memory for better network performance.

You can find the full source code on GitHub including a SAM template to provision the AWS resources. Have fun!


This is a shorter article. Do you prefer longer or shorter reads? Let me know! michael@widdix.de, LinkedIn, or @hellomichibye.

Michael Wittig

Michael Wittig

I’m the author of Amazon Web Services in Action. I work as a software engineer, and independent consultant focused on AWS and DevOps.

You can contact me via Email, Twitter, and LinkedIn.

Briefcase icon
Hire me
Cover of Rapid Docker on AWS

New book: Rapid Docker on AWS

A rapid way to get your web application up and running on AWS. Made for web developers and DevOps engineers who want to dockerize their web applications and run their containers on Amazon Web Services. Prior knowledge of Docker and AWS is not required.

Buy icon
Buy now
Marbot Logo

Incident Management for Slack

Team up to solve incidents with our chatbot marbot. Never miss a critical alert. Escalate alerts from your AWS infrastructure among your team members. Strong integrations with all parts of your AWS infrastructure: CloudWatch, Elastic Beanstalk, RDS, EC2, ...

Slack icon
Try for free
📚 Rapid Docker on AWS
A rapid way to get your web application up and running on AWS. Learn how to package your application into Docker containers. Learn more.